拟牛顿迭代方法初始值的研究  

The Initial Value Research Based on Quasi-Newton Method

作　　者：张安玲[1] ZHANG Anling(Department of Mathematics,Changzhi University,Changzhi 046011,China)

机构地区：[1]长治学院数学系,山西长治046011

出　　处：《太原师范学院学报（自然科学版）》2018年第2期6-8,共3页Journal of Taiyuan Normal University:Natural Science Edition

基　　金：山西省高等教学改革创新项目(J2018180)

摘　　要：针对拟牛顿法对初始值敏感的问题,提出一种粒子群优化算法和拟牛顿法相结合的方法.该方法首先利用粒子群优化算法的全局搜索性对所求问题在可行解区域范围内进行大范围的搜索,搜索到一定程度,把当代的最好点作为拟牛顿法的初始值进行拟牛顿法迭代.数值结果表明,该方法有效地解决了拟牛顿法对初始值的敏感性问题,保证拟牛顿法的收敛性.Aiming at the problems of quasi-Newton method such as high sensitivity to the initial value,a combined method of particle swarm optimization algorithm and quasi-Newton method is proposed.This method carries out a wide range of search in the range of feasible solution region using global searching capability of particle swarm optimization algorithm.To a certain extent,the contemporary best points are used to the initial value of the quasi-Newton method.Quasi-Newton method iteration search based on these initial values is adopted.The numerical results show that this method can effectively solve the quasi-Newton method problem which is sensitive to the initial value.At the same time,it will guarantee the convergence of quasi-Newton method.

关 键 词：拟牛顿法 初始值 粒子群优化算法 最优解 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]